Artificial intelligence has long been used in theoretical scientific research, even with some success in mathematics and materials science, but research also involves practical experiments. In this area, scientists have turned to a new human solution called the Model Hardware Standard.
Image source: Human
As mentioned before, this solution allows financial timesto manage laboratory equipment with software interfaces. Electron microscopes, fluid delivery systems, and robotic arms can now operate autonomously, performing actual physical, chemical, and biological experiments at the behest of artificial intelligence. Anthropic has already tested its development in a small group of science labs, including those involved in quantum computing. According to participants in these experiments, in this case the AI agent replaced the live lab assistant quite successfully.
For example, a system based on Crowder’s model was able to independently focus an electron microscope on a desired region of human brain tissue at the command of a scientist, even though no one had previously taught it the location of that region. Anthropic intends to distribute its model for managing scientific experiments as open source code, but before making it available, it must provide mechanisms to prevent errors. CEO Dario Amodei is also concerned about the threat of using artificial intelligence to create biological weapons and toxic substances.
Artificial intelligence is already helping pharmaceutical companies find new drugs. Nvidia, together with specialist Eli Lilly, is providing financial support to Insilico Medicine, a new startup specializing in this field. For example, Demis Hassabis, founder of Google DeepMind, moved to work for a new startup called Isomorphic Labs, which also uses artificial intelligence to develop drugs. In 2024, he will win the Nobel Prize in Chemistry for his progress in using artificial intelligence to predict protein structures.
If you find an error, select it with your mouse and press CTRL+ENTER.
